Answer:
a) Ub = 0,069
b) CI 99 % = ( 0,0287 ; 0,069 )
c) n = 2536
Explanation:
Confidence Interval (CI) for proportion is
p₀ ± z(c) * √ ( p₀*q₀ ) /n
Where p₀ is the sample proportion
n = size of the sample
and z(c) is z score for the significance level α
As CI = 99 % α = 1% α = 0,01 α/2 = 0,005
from z-table we find z(c) = 2,57
p₀ = 12/300 p₀ = 0,04 ;
n*p₀ = 12 > 10 and n*q₀ = 300*0,96 n*q₀ = 288 > 10
We can use normal distribution as a valid aproximation of binomial distibution
Then z(c) = 2,57
a) upper bound for CI Ub
Ub = p₀ + z(c) * √ p₀q₀/300
Ub = 0,04 + 2,57 * √ (0,04*0,96)/300
Ub = 0,04 + 0,0113
Ub = 0,069
b) CI at 99% is
CI = p₀ ± z(c) * √ ( p₀*q₀ ) /n
we have already calculated the Ub = 0,069
the Lower bound would be Lb = 0,04 - 0,0113
Lb = 0,0287
Then CI 99 % = ( 0,0287 ; 0,069 )
c) If CI ≤ 0,02 we can find the minimun n value with the equation:
p₀ ± z(c) * √ ( p₀*q₀ ) /n = CI or p₀ - z(c) * √ ( p₀*q₀ ) /n = 0,01
For symmetry respect to p₀
z(c) *√ (p₀*q₀) / n = 0,01
Solving for n
z(c) * √ ( p₀*q₀ ) /n = 0,01
2,57 * √ (0,04*0,96)/n = 0,01
2,57 * √ 0,0384/n = 0,01
2,57 = 0,01/√ 0,0384/n
Squaring both sides of the equation
6,6049 = (0,01)² *n / 0,0384
0,0384*6,6049 / 0,0001 = n
n = 2536